Method and system to generate persona-based commentary data for machine learning model card document

Abstract

This disclosure relates generally to method and system to generate persona-based commentary data for machine learning model card document. Existing techniques on model card are designed mainly for personas and understanding section of the model card document requires a certain level of expertise in machine learning. The method of the present disclosure receives a model card document comprising a plurality of sections and the model card document corresponds to a persona. Each section of the model card document obtains a metadata for the persona. The data curator machine learning model automatically generates a persona-based report trajectory for a plurality of sections of the metadata and a plurality of schema rules to generate a prompt template. The commentary generator ML model generates one or more commentary data for each section associated with the prompt template corresponding to the persona.

What is claimed is: 
     
         1 . A processor-implemented method to generate persona based commentary data, comprising:
 receiving via one or more hardware processor, a model card document comprising a plurality of sections corresponding to a persona, wherein the persona represents a type of person with role specific domain communication characteristics;   obtaining for each section of the model card document via the one or more hardware processors, a metadata from a metadata store for the persona, wherein the metadata store comprises at least one of (i) a structured metadata and (ii) an unstructured metadata;   preprocessing via the one or more hardware processors, the metadata to obtain relevant information for the plurality of sections of the model card document;   feeding the preprocessed metadata into a data curator machine learning (ML) model via the one or more hardware processors, to obtain information associated with a plurality of sections corresponding to the persona;   generating by the data curator ML model via the one or more hardware processors, a persona-based report trajectory for each section, based on the metadata of corresponding section and a plurality of schema rules, wherein a prompt template is generated for the persona-based report trajectory; and   generating by a commentary generator machine learning (ML) model via the one or more hardware processors, one or more commentary data for each section associated with each prompt template corresponding to the persona.   
     
     
         2 . The processor-implemented method of  claim 1 , wherein each prompt template is generated based on each persona-based report trajectory for the corresponding section of the metadata. 
     
     
         3 . The processor-implemented method of  claim 1 , wherein the commentary generator machine learning model is trained to generate one or more commentary data for the report trajectory corresponding to the persona by performing the steps of:
 providing a training dataset to identify each section of the metadata corresponding to each persona on writing domain-aware content and use of language in commentary; and   learning by the commentary generator machine learning model to generate the one or more commentary data for each section corresponding to the persona.   
     
     
         4 . The processor-implemented method of  claim 1 , wherein the plurality of sections of the metadata comprises data required for each section of the model card document of the corresponding persona. 
     
     
         5 . The processor-implemented method of  claim 1 , wherein the persona-based report trajectory comprises a plurality of trajectory parameters and its associated values. 
     
     
         6 . The processor-implemented method of  claim 1 , wherein the plurality of schema rules comprises rules to generate the one or more commentary data, a language content, domain, and at least one language requirement of the persona for each section. 
     
     
         7 . The processor-implemented method of  claim 1 , wherein each prompt template is optimized using at least one of an iterative rephrasing technique and an iterative evaluation technique. 
     
     
         8 . The processor-implemented method of  claim 1 , wherein the plurality of schema rules relates to each rule comprising eligible personas, a list of data regular expressions, a readability level, and a list of domains. 
     
     
         9 . A system, to generate persona based commentary data comprising:
 a memory storing instructions;   one or more communication interfaces; and   one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to:   receive a model card document comprising a plurality of sections and the model card document corresponds to a persona, wherein the persona represents a type of person with role specific domain communication characteristics;   obtain for each section of the model card document, a metadata from a metadata store for the persona, wherein the metadata store comprises at least one of (i) a structured metadata and (ii) an unstructured metadata;   preprocess the metadata to obtain relevant information of the plurality of sections of the model card document;   feed the preprocessed metadata into a data curator machine learning (ML) model to obtain information associated with a plurality of sections corresponding to the persona;   generating by the data curator ML model a persona-based report trajectory for each section, based on the metadata of corresponding section and a plurality of schema rules, wherein a prompt template is generated for the persona-based report trajectory; and   generate by a commentary generator machine learning (ML) model one or more commentary data for each section associated with the prompt template corresponding to the persona.   
     
     
         10 . The system of  claim 9 , wherein the prompt template is generated based on each persona-based report trajectory for the corresponding section of the metadata. 
     
     
         11 . The system of  claim 9 , wherein the commentary generator machine learning model is trained to generate one or more commentary data for the report trajectory corresponding to the persona by performing the steps of:
 providing a training commentary dataset to identify each section of the metadata corresponding to each persona on writing domain-aware content and use of language in commentary; and   learning by the commentary generator machine learning model to generate the one or more commentary data for each section corresponding to the persona.   
     
     
         12 . The system of  claim 9 , wherein the plurality of sections of the metadata comprises data required for each section of the model card document of the corresponding persona. 
     
     
         13 . The system of  claim 9 , wherein the persona-based report trajectory comprises a plurality of trajectory parameters and its associated values. 
     
     
         14 . The system of  claim 9 , wherein the plurality of schema rules comprises rules to generate the one or more commentary data, a language content, domain, and at least one language requirement of the persona for each section. 
     
     
         15 . The system of  claim 9 , wherein each prompt template is optimized using at least one of an iterative rephrasing technique and an iterative evaluation technique. 
     
     
         16 . The system of  claim 9 , wherein the plurality of schema rules relates to each rule comprising eligible personas, a list of data regular expressions, a readability level, and a list of domains. 
     
     
         17 . One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:
 receiving a model card document comprising a plurality of sections corresponding to a persona, wherein the persona represents a type of person with role specific domain communication characteristics;   obtaining for each section of the model card document a metadata from a metadata store for the persona, wherein the metadata store comprises at least one of (i) a structured metadata and (ii) an unstructured metadata;   preprocessing the metadata to obtain relevant information for the plurality of sections of the model card document;   feeding the preprocessed metadata into a data curator machine learning (ML) model to obtain information associated with a plurality of sections corresponding to the persona;   generating by the data curator ML model a persona-based report trajectory for each section based on the metadata of corresponding section and a plurality of schema rules, wherein a prompt template is generated for the persona-based report trajectory; and   generating by a commentary generator machine learning (ML) model one or more commentary data for each section associated with each prompt template corresponding to the persona.   
     
     
         18 . The one or more non-transitory machine-readable information storage mediums of  claim 17 , wherein each prompt template is generated based on each persona-based report trajectory for the corresponding section of the metadata. 
     
     
         19 . The one or more non-transitory machine-readable information storage mediums of  claim 17 , wherein the commentary generator machine learning model is trained to generate one or more commentary data for the report trajectory corresponding to the persona by performing the steps of:
 providing a training dataset to identify each section of the metadata corresponding to each persona on writing domain-aware content and use of language in commentary; and   learning by the commentary generator machine learning model to generate the one or more commentary data for each section corresponding to the persona.   
     
     
         20 . The one or more non-transitory machine-readable information storage mediums of  claim 17 , wherein the plurality of sections of the metadata comprises data required for each section of the model card document of the corresponding persona.